1. Field of the Invention
The present invention relates generally to the field of semiconductor processing, and more particularly, to a method of fabricating a feature in an integrated circuit.
2. Description of the Related Art
A desire to increase the speed and density of integrated circuits (ICs) has led to progressive reductions in feature dimensions, particularly in the lengths of metal-oxide semiconductor (MOS) transistor gate electrodes. However, the reduction in feature dimensions is limited by a need to provide some margin for process variation, which is responsible for a difference between the target dimension and the minimum actual dimension typically yielded by the fabrication process. For example, given the process variation of conventional photolithography, it is difficult to manufacture MOS transistor gate electrodes with a target dimension of less than 150 nm. To overcome this difficulty, a process of gate definition referred to as Spacer Gate (SG) has been developed.
An example of an SG process flow on a semiconductor wafer is illustrated in FIGS. 1a through 1g, each of which represent a cross sectional view of the wafer. FIG. 1a shows a silicon substrate 101 with a gate oxide layer 102 formed thereon. A polysilicon layer 103 has then been formed on the gate oxide layer 102. The gate electrode of the future MOS transistor will be formed from the polysilicon layer 103.
FIG. 1b shows an edge definition area 104 of silicon dioxide that has been formed by patterning a layer of silicon dioxide using conventional photolithography and etching processes. The patterning of edge definition area 104 constitutes the first masking step of this SG process. FIG. 1c shows a silicon nitride spacer layer 105 formed on the wafer. FIG. 1d shows silicon nitride spacers 106 and 107 that have been formed on the edge of edge definition area 104 by an anisotropic each of spacer layer 105. FIG. 1e shows the wafer after edge definition area 104 has been etched away.
In FIG. 1e, spacers 106 and 107 remain on the wafer, as part of a continuous ring of silicon nitride that was formed around the entire edge of edge definition area 104. Since the area of gate electrode layer 103 that is covered by silicon nitride will not be subsequently removed and the desired pattern of the gate electrode is not a continuous ring, a portion of the silicon nitride ring must be removed. The removal of a portion of the silicon nitride ring is referred to as a nitride trim, which constitutes the second masking step of this SG process. In FIG. 1e, spacer 106 covers an area of polysilicon layer 103 that will become a portion of a gate electrode, but spacer 107 represents a portion of the spacer ring that must be trimmed. FIG. 1f shows spacer 106 covered by a photoresist trim mask 108, which is formed by conventional photolithography.
FIG. 1g shows the wafer after spacer 107 has been removed by a trim etch and trim mask 108 has been stripped. The area of polysilicon layer 103 that is under spacer 106 will become a gate electrode with a length that depends on the length 109 of spacer 106. The length 109 of spacer 106 depends on the thickness of former spacer layer 105, so only features of the same length as the future gate electrode can be defined with this SG process. Therefore, a third masking step is needed to define features from polysilicon layer 103 that are of greater length than the future gate electrode. Such features might include transistor gate electrodes of greater than minimum length, polysilicon interconnect lines, and polysilicon contact pads. The photoresist mask used for this third masking step is referred to as a pad mask.
FIG. 1h shows pad mask 110 masking an area of polysilicon layer 103 that will be protected during polysilicon etch to form a contact pad. FIG. 1l shows gate electrode 111 and contact pad 112 after polysilicon etch. Next, pad mask 110 is stripped. Seal oxide 113, shown in FIG. 1j, is grown to protect the edges of gate electrode 111 and contact pad 112 during subsequent processing. Finally, FIG. 1k shows the wafer after spacer 106 has been etched away. The remainder of the transistor structure can be formed using conventional MOS processing.
The above-described method of fabricating a gate electrode using a spacer requires three masking steps. The relative high cost of processing and the long throughput time associated with the three-mask process make it difficult to integrate the three-mask process into a main stream MOS process. Further, process complexity and cost are directly related to the number of masking steps. Therefore, a method of fabricating a feature using a spacer that requires only two masking steps is desired.